Broadside random access memory for low cycle memory access and additional functions

Abstract

A computational system includes one or more processors. Each processor has multiple registers, as well attached memory to hold instructions. The processor is coupled to one or more broadside interfaces. A broadside interface allows the processor to load or store an entire widget state in a single clock cycle of the processor. The broadside interface also allows the processor to move and store 32 bytes of information into RAM in less than four to five clock cycles of the processor while the processor concurrently performs one or more mathematical operations on the information while the move and store operation is taking place.

What is claimed is: 
     
       1. A computational device, comprising:
 a processor, wherein: 
 the processor has an instruction set that has a first bit width, 
 the processor includes a processor clock and a register, an interface coupled to the register; and 
 a random-access memory (RAM) coupled to the interface, wherein: 
 the RAM has a second bit width that is greater than the first bit width: 
 the interface is configured to:
 store or retrieve accelerator state data from the register in a single clock cycle of the processor clock; and 
 
 store or retrieve a set of data having the second bit width from the RAM within three clock cycles of the processor clock from when the processor executes a corresponding memory operation; and 
 the processor is configured to perform one or more mathematical operations on at least one portion of the set of data during execution of the corresponding memory operation. 
 
     
     
       2. The computational device of  claim 1 , further comprising:
 a glue logic coupled to the processor, and wherein the processor is further configured to store an address of the RAM in the glue logic, the address corresponding to an address most recently accessed by the processor. 
 
     
     
       3. The computational device of  claim 1 , further comprising:
 a checksum circuit, and 
 wherein the processor is further configured to perform the one or more mathematical operations on the at least one portion of the set of data during execution of the corresponding memory operation by performing a checksum operation on the at least one portion of the set of data during the execution of the corresponding memory operation using the checksum circuit. 
 
     
     
       4. The computational device of  claim 1 , wherein the processor is further configured to perform an endian flipping operation on the at least one portion of the set of data during the execution of the corresponding memory operation. 
     
     
       5. The computational device of  claim 1 , wherein the processor is further configured to transpose a first content of the register with a second content of the register. 
     
     
       6. The computational device of  claim 1 , wherein the processor is further configured to implement a queue data structure using the RAM in an address auto-increment mode. 
     
     
       7. The computational device of  claim 1 , wherein the processor is further configured to implement a first-in-first-out data structure using the RAM in an address auto-increment mode. 
     
     
       8. The computational device of  claim 1 , wherein the processor is further configured to implement a last-in-first-out data structure using the RAM in an address auto-increment mode. 
     
     
       9. The computational device of  claim 1 , further comprising a filter database connected to the RAM, and wherein the RAM is further configured to perform look up operations using the filter database. 
     
     
       10. The computational device of  claim 1 , wherein the first bit width is 32 bits and the second bit width is 32 bytes. 
     
     
       11. A computer-implemented method, comprising:
 loading, using an interface, accelerator state data to a register in a single clock cycle of a processor coupled to the register; 
 executing a memory operation using the processor, wherein the executing of the memory operation comprises loading or storing a set of data to a RAM coupled to the interface, in less than four clock cycles of the processor, wherein: 
 the processor has an instruction set haying a first bit width; and 
 the RAM has a second bit width that is greater than the first bit width; and 
 performing, using the processor, one or more mathematical operations on at least one portion of the set of data during the execution of the memory operation. 
 
     
     
       12. The computer-implemented method of  claim 11 , further comprising:
 performing, using the processor, a checksum operation on the at least one portion of the set of data during the execution of the memory operation using a checksum hardware circuit. 
 
     
     
       13. The computer-implemented method of  claim 11 , further comprising performing, using the processor, an endian flipping operation on the at least one portion of the set of data during the execution of the memory operation. 
     
     
       14. The computer-implemented method of  claim 11 , further comprising, transposing, by the processor, a first content of a first plurality of registers coupled to the processor, with a second content of a second plurality of registers coupled to the processor. 
     
     
       15. The computer-implemented method of  claim 11 , further comprising implementing, by the processor, a queue data structure by using the RAM in an address auto-increment mode. 
     
     
       16. The computer-implemented method of  claim 11 , further comprising implementing, by the processor, a FIFO data structure by using the RAM in an address auto-increment mode. 
     
     
       17. A system on chip, comprising:
 a programmable real-time unit, wherein: 
 the programmable real-time unit is responsive to an instruction set having a first bit width: and 
 the programmable real-time unit includes a register; 
 an interface circuit coupled to the register; and 
 a random-access memory (RAM) coupled to the interface circuit, 
 wherein the interface circuit is configured to load or store accelerator state data to the register in a single clock cycle of the programmable real-time unit; 
 the interface circuit is further configured to load or store a set of data having a second bit width that is greater than the first bit width to the RAM within one clock cycle of the programmable real-time unit when the programmable real-time unit executes a memory operation; and 
 wherein the programmable real-time unit is configured to perform one or more mathematical operations on at least one portion of the set of data during execution of the memory operation. 
 
     
     
       18. The system on chip of  claim 17 , further comprising:
 a glue logic circuit coupled to the programmable real-time unit, and wherein the programmable real-time unit is further configured to store an address of the RAM using the glue logic circuit, the address corresponding to an address most recently accessed by the programmable real-time unit. 
 
     
     
       19. The system on chip of  claim 17 , further comprising:
 a checksum circuit, and 
 wherein the programmable real-time unit is further configured to perform the one or more mathematical operations on the at least one portion of the set of data during the execution of the memory operation by performing a checksum operation on the at least one portion of the set of data during the execution of the memory operation using the checksum circuit. 
 
     
     
       20. The system on chip of  claim 17 , wherein the first bit width is 32 bits and the second bit width is 32 bytes.